Speaker Diarization Script
by pyannote · open-source · Last verified 2026-03-17
Identifies and segments individual speakers in multi-party audio recordings using pyannote.audio, then merges speaker labels with Whisper transcriptions to produce a turn-by-turn conversation transcript. Supports overlapping speech detection and speaker enrollment for known identities.
https://github.com/pyannote/pyannote-audio ↗B
B—Above Average
Adoption: B+Quality: AFreshness: ACitations: BEngagement: F
Specifications
- License
- MIT
- Pricing
- open-source
- Capabilities
- speaker-segmentation, overlap-detection, whisper-merge, speaker-enrollment
- Integrations
- pyannote-audio, whisper, torch, huggingface
- Use Cases
- interview-transcription, legal-deposition-analysis, conference-call-notes
- API Available
- No
- Language
- python
- Dependencies
- pyannote.audio, openai-whisper, torch, torchaudio, huggingface-hub
- Environment
- Python 3.10+, CUDA recommended
- Est. Runtime
- 5-15 minutes per hour of audio
- Tags
- diarization, speaker-id, pyannote, audio-segmentation, multi-speaker
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
60.4Adoption
72
Quality
83
Freshness
85
Citations
60
Engagement
0